CRISPR-Cas9 Screening and Simulated Infection Transcriptomic Identify Key Drivers of Innate Immunity in Bactrian Camels

Simple Summary

Bactrian camels are animals known for surviving in harsh environments, with unique physiological and immune adaptations. This study aimed to understand how they defend against viral and bacterial infections. By combining gene-editing technology with gene expression analysis, the researchers systematically identified key genes important for fighting viruses and bacteria. The results showed that the genetic mechanisms of antiviral and antibacterial activity are not entirely the same. Key genes such as HSP90AA1 and CSF1 were found to play central roles in immune defense. These findings improve our understanding of camel immunology and provide new insights into how mammals adapt immune systems to infections. In the future, this knowledge could help breed more disease-resistant livestock and support sustainable animal farming.

Abstract

The Bactrian camel (Camelus bactrianus), with its unique physiological adaptations and immune characteristics, represents a highly valuable model for innate immunity research. However, a systematic dissection of its innate immune gene repertoire and the key functional drivers within its immune response remains limited. This study integrated CRISPR-Cas9 knockout screening with time-resolved transcriptomic profiling to systematically unveil the immune regulatory mechanisms in camel dermal fibroblasts challenged with the viral mimic poly(I:C) and the bacterial mimic LPS. The CRISPR screen successfully identified 59 key genes conferring a survival advantage under lethal pathogenic challenge. The gene sets required for resisting viral versus bacterial mimics were entirely distinct, revealing divergent genetic underpinnings. Transcriptomic analysis further delineated a dynamic reprogramming of gene expression, uncovering a shared core immune response program alongside significant stimulus-specific regulation. Integrative analysis pinpointed pivotal genes, such as HSP90AA1 in the antiviral process and CSF1 in the antibacterial process, which played critical roles at both the functional screening and transcriptional regulatory levels. These key genes exhibited dynamic and evolving co-expression networks across different time points, indicating their temporally specific regulatory roles throughout the immune response. By combining functional genomics and transcriptomics, this study provides the first systematic mapping of the innate immune landscape and its dynamic regulation in the Bactrian camel, not only deepening the understanding of camelid immunobiology but also offering a new framework and insights for evolutionary studies of immune adaptation mechanisms in mammals.

1. Introduction

The significance of immunological research on Bactrian camels is not only reflected in their resilience as economic livestock but also in their value as a mammalian model organism [1,2,3,4,5,6]. Bactrian camels are a crucial source of milk, meat, and wool in the cold and desert regions of Central Asia, playing a vital role in the livelihoods and economies of these areas [7,8,9]. On the other hand, camels, recognized for their production of heavy-chain antibodies [10,11] and extraordinary adaptability to harsh environments [1], demonstrate unique genomic characteristics in their immune response [2,4,12,13]. This combination of unique attributes emphasizes the specialized evolution of their immune system, positioning them as a significant mammalian model in the field of immunological research.

As a principal arm of the immune system, innate immunity provides the initial line of defense against a broad range of pathogens. In humans, it also serves as a critical lens through which to observe the evolutionary pressures exerted by pathogenic and commensal microorganisms on the host genome [14,15]. Research into innate immunity has expanded across diverse species, identifying key genes in models ranging from humans [16,17] to oysters [18], and insects [19,20]. In recent years, pooled and arrayed CRISPR-based genetic screens have revolutionized our understanding of immune regulation. Such screens have unveiled previously unrecognized intracellular drivers of both innate and adaptive immunity, leading to insights into host-pathogen interactions [21], immune cell differentiation [22], and regulatory networks [23]. The transformative role of CRISPR screens in immunology has been systematically reviewed elsewhere [24,25]. Genetic screening provides a powerful approach for identifying genes, pathways and mechanisms involved in a given phenotype or biological process. In human cellular and disease models, CRISPR-based knockout screens have made significant strides across major research areas, including the identification of key host factors for antiviral defense [26,27], the discovery of novel targets for cancer therapy [28], the elucidation of mechanisms underlying drug resistance [29], and the dissection of disease pathogenesis [30,31,32]. Beyond human biomedicine, this technology has also been widely applied to enhance economically important traits in livestock, such as improving productivity, disease resistance, and environmental adaptability. Notable examples include editing growth-related genes to improve yields, knocking in disease-resistance genes (e.g., NRAMP1 in cattle) or knocking out viral receptor genes (e.g., CD163 in pigs) to breed resistant lines, and employing genome-wide CRISPR screening to identify viral host factors, thereby providing novel targets and strategies for disease-resistance breeding [33,34,35,36].

Despite these advances, the molecular mechanisms underpinning the innate immune responses of the Bactrian camel remain largely unexplored. In this study, we leverage a focused CRISPR-Cas9 knockout library targeting innate immunity-related genes in Bactrian camel fibroblasts, coupled with selection under viral (poly(I:C)) and bacterial (LPS) mimetics. Through this approach, we aim to identify key genes that confer a survival advantage during pathogenic challenges. Moreover, we integrate this genetic screening with time-series transcriptomic analyses to capture the dynamic landscape of gene expression in response to viral and bacterial stimuli. By combining functional genetic approaches with transcriptional profiling, this study not only highlights critical determinants of camel innate immunity but also provides a comprehensive temporal perspective, advancing our understanding of the cellular defense mechanisms deployed during pathogenic encounters.

2. Methods

2.1. Compilation of the Camel Innate Immune Gene Set

To define the genetic repertoire for screening, we first curated a comprehensive list of human innate immunity genes by integrating data from Gene Ontology (GO) [37], InnateDB [38], and a key reference study [17]. Using these human genes as queries, we performed a comparative genomic analysis against the domestic Bactrian camel genome (Assembly: Ca_bactrianus_MBC_1.0) using OrthoVenn3 [39]. The OrthoMCL algorithm was applied with an E-value cutoff of 1 × 10⁻² to identify high-confidence orthologous genes, resulting in a final set of 1535 camel innate immunity genes (CIIGs) for downstream library design.

2.2. Primary Cell Culture and Experimental Strategy

Bactrian camel dermal fibroblast (CDF) cells were isolated from the ear tip tissue of a single, healthy six-year-old female using the explant culture method. To maintain phenotypic stability, all experiments were conducted with low-passage (P3) cells that were 80–90% confluent and in the logarithmic growth phase. Cells were maintained in DMEM/F12 medium supplemented with 10% fetal bovine serum (FBS, HyClone, Logan, UT, USA) and 1% penicillin/streptomycin (Thermo Fisher Scientific, Waltham, MA, USA) at 37 °C under 5% CO₂ to prevent bacterial contamination during routine culture. Prior to all stimulation experiments, cells were washed with phosphate-buffered saline (PBS) and cultured in antibiotic-free medium for at least 24 h to eliminate potential effects of antibiotics on immune responses. The immunostimulants used were poly(I:C) (Invivogen, Toulouse, France, Cat. No. tlrl-pic) and LPS (Sigma-Aldrich, St. Louis, MO, USA, Cat. No. L4524) (extracellular delivery), which were reconstituted according to the manufacturers’ instructions. Our functional genomics pipeline consisted of three sequential phases: (1) determining the cytotoxic concentration of immune stimulants via a viability assay; (2) conducting a CRISPR/Cas9 knockout screen to identify genes essential for survival under immune challenge; and (3) performing time-resolved transcriptomic profiling to elucidate the underlying molecular pathways. All animal procedures were approved by the Animal Care and Use Committee of Inner Mongolia Agricultural University (Approval No. NND2023110).

2.3. Cell Viability Assay

Cell viability was assessed using a Cell Counting Kit-8 (CCK-8) assay. Briefly, CDF cells were seeded in 96-well plates (1 × 10⁴ cells/well; n = 3 technical replicates) and treated for 24 h with a gradient of concentrations (0, 1, 5, 10, 20, 50 µg/mL) of poly(I:C) or LPS. After treatment, cells were incubated with 10 µL of CCK-8 reagent for 2 h at 37 °C, and the absorbance at 450 nm was measured using a microplate reader (BioTek Synergy H1, Winooski, VT, USA) to assess cell viability.

2.4. CRISPR Library Construction and Screening

A custom sgRNA library was constructed to target 1535 identified innate immune genes. For each gene, we designed three to seven sgRNAs according to stringent selection criteria [40]. The 20-nucleotide guide sequences were preferentially targeted to conserved protein-coding exons, with 83% situated within the first two exons to maximize the probability of functional knockout. All sgRNAs were designed to maintain a GC content of 40–60% and to avoid poly-T stretches exceeding four consecutive thymines. To minimize off-target effects, we employed a two-tiered strategy: initial filtering for seed region specificity (allowing ≤3 mismatches in nucleotides 1–12) was followed by application of the computational Cutting Frequency Determination (CFD) score with a stringent threshold of <0.2. The sgRNAcas9 (v3.0.5) software [41] was used to search for highly specific sgRNA candidates. Additionally, a set of 1000 non-targeting control sgRNAs was designed in silico. These control sequences were rigorously screened to ensure the absence of any perfect match or near-perfect match (≤3 mismatches) to the reference genome, thereby serving as reliable negative controls for background normalization and false discovery rate (FDR) calculation in downstream analyses. The final optimized library comprised 11,741 sgRNA constructs, including 10,741 targeting sgRNAs (covering 1535 CIIGs) and 1000 non-targeting negative controls for screening validation (see Table S2 for details).

Library oligonucleotides (Syno^®3.0, GenScript, Piscataway, NJ, USA) were PCR-amplified using Phusion High-Fidelity DNA Polymerase and primers that appended the necessary adapter sequences, including homology arms for Gibson Assembly, following an established protocol [42]. The resulting amplicons, which contained sgRNA expression cassettes flanked by vector-compatible overhangs, were cloned into the BsmBI-digested lentiGuide-Puro vector (Addgene, Watertown, MA, USA, #52963) via Gibson Assembly. The assembled plasmid library was then transformed into Endura™ electrocompetent E. coli (Lucigen, Middleton, WI, USA, Cat. No. 60242-2), and the sgRNA plasmid library was extracted on a large scale using the PureLink HiPure Plasmid Filter Maxiprep Kit (Thermo Fisher Scientific). To validate the quality and sequence integrity of the synthesized library, the plasmid pool was subjected to deep sequencing. The plasmid pool was subjected to deep sequencing to validate library quality and sequence integrity. The analysis confirmed excellent coverage and even representation of all sgRNAs.

Lentiviral particles were produced by co-transfecting 293T cells (ATCC, Manassas, VA, USA, Cat. No. CRL-3216) and concentrated by polyethylene glycol (PEG) precipitation. The functional viral titer reached 1.2 × 10⁸ TU/mL as determined by puromycin resistance assay. To enable screening, a Cas9-expressing polyclonal CDF-1-Cas9 cell line was first established by lentiviral transduction followed by blasticidin selection. For the genetic screen, CDF-1-Cas9 cells were transduced with the lentiviral sgRNA library at a low multiplicity of infection (MOI = 0.3) to ensure single sgRNA incorporation per cell. Forty-eight hours post-transduction, cells were selected with 2 μg/mL puromycin (Sigma-Aldrich, Cat. No. P8833) for 7 days to eliminate non-transduced cells. The resulting polyclonal sgRNA library cell pool was divided into two independent screening groups and one untreated control group. Each screening group underwent two rounds of positive selection under a specific immune stimulus: one group was challenged with 50 µg/mL poly(I:C) to mimic viral infection, and the other with 50 µg/mL LPS to simulate bacterial infection. Each treatment round consisted of a 24 h stimulus exposure followed by a 48 h recovery period. This selection regimen induced >90% cell death in each challenged group, thereby stringently enriching for knockout clones that conferred a survival advantage under the specific immune challenge.

Genomic DNA was extracted from the surviving cell population using the DNeasy Blood & Tissue Kit (Qiagen, Valencia, CA, USA, Cat. No. 69504). The integrated sgRNA sequences were recovered by PCR amplification with specific primers and subsequently sequenced on an Illumina NovaSeq platform. The enrichment of specific sgRNAs in the post-selection population compared to the initial library was quantitatively analyzed using the MAGeCK algorithm [43], and genes associated with significantly enriched or depleted sgRNAs (FDR < 0.05) were identified as high-confidence hits.

2.5. Time-Course Transcriptomic Analysis of Camel Fibroblasts

CDF cells were treated with a signaling-active concentration of 10 µg/mL poly(I:C) or LPS and harvested at 0-, 6-, 12-, and 24 h post-treatment (n = 3 technical replicates per group). Total RNA was extracted with TRIzol Reagent (Thermo Fisher, Cat. No. 15596026). RNA integrity was verified (RIN > 8.0) using an Agilent 2100 Bioanalyzer with the RNA Nano 6000 Assay Kit (Agilent, Santa Clara, CA, USA, Cat. No. 5067-1511). Sequencing libraries were prepared from 1 µg of total RNA per sample using the NEBNext Ultra II Directional RNA Library Prep Kit (New England Biolabs, Ipswich, MA, USA). Briefly, mRNA was enriched using oligo(dT) beads and then fragmented. First-strand cDNA synthesis was performed with random hexamer primers and ProtoScript II Reverse Transcriptase, followed by second-strand synthesis using DNA Polymerase I and RNase H. The double-stranded cDNA fragments were end-repaired, adenylated at their 3′ ends, and ligated to Illumina sequencing adapters. The final libraries were amplified with PCR, quantified, and sequenced on an Illumina NovaSeq 6000 platform in paired-end mode with 150 bp read length.

The quality of the raw sequencing data was first assessed using FastQC (v0.11.9) to evaluate metrics including per-base quality scores, nucleotide composition, GC content, and sequence duplication levels. Subsequently, fastp (v0.23.0) was employed for quality control and adapter trimming. The filtering criteria were as follows: removal of adapter sequences, discarding of reads where over 20% of bases had a Phred quality score below 20 (Q20), and elimination of reads containing more than 10% undefined bases (N). This process yielded high-quality clean reads for subsequent analysis.

The clean reads were then aligned to the domestic Bactrian camel reference genome using HISAT2 (v2.0.0) [44], a memory-efficient and rapid alignment tool based on the Burrows-Wheeler Transform and Ferragina-Manzini index. The alignment was performed with default parameters. The resulting SAM files were sorted and converted to BAM format using SAMtools (v1.9). Transcript assembly and quantification of gene-level abundances were performed from the sorted BAM files using StringTie (v1.3.3). Gene expression levels were quantified as both raw read counts and TPM (Transcripts Per Million) to cater to different downstream analytical requirements.

To assess sample reproducibility and identify potential outliers, Principal Component Analysis (PCA) was performed on the variance-stabilized transformed count data using the DESeq2 [45] package. For the time-series transcriptomic data, differential expression analysis was conducted using a sequential comparison strategy to capture dynamic gene regulation. Specifically, for each stimulus condition, the transcriptome at each post-stimulation time point (6, 12, 24 h) was compared against its immediately preceding time point (e.g., 6 h vs. 0 h, 12 h vs. 6 h, 24 h vs. 12 h). All differential expression analysis between conditions was conducted using DESeq2 (v1.40.0), applying the Wald test. Genes with an adjusted p-value (Benjamini-Hochberg FDR) < 0.05 and an absolute log₂ fold change >1 were defined as significantly differentially expressed. Subsequent data analysis, visualization (including correlation, scatter plots, and heatmaps), and statistical computation were performed in the R environment [46].

Temporal expression patterns of differentially expressed genes were analyzed using Fuzzy C-Means (FCM) clustering implemented in the Mfuzz package (v2.32.0) [47] in R (v4.4.1). The clustering was performed with a fuzzification coefficient (M) of 1.7, and the optimal number of clusters (k = 6) was determined through evaluation of within-cluster sum of squares (elbow method), enforcement of a minimum cluster size (>50 genes), and assessment of biological relevance. The algorithm’s convergence criteria were set at either 100 maximum iterations or an objective function change (ΔJ) threshold of 1 × 10⁵. The visualization of network graphs was performed using Cytoscape (v3.9.1) [48]. For gene function and pathway enrichment analysis, we utilized the g: Profiler [49] with g:SCS significance threshold (p < 0.05) to obtain the latest gene annotation of pathways.

2.6. Statistical Analysis

Statistical analyses were performed as detailed in respective method subsections and are summarized here. Cell viability data (mean ± SD of n = 3 technical replicates) were analyzed using the non-parametric Kruskal–Wallis test, followed by Dunn’s post hoc test for multiple comparisons. The CRISPR screen was analyzed using MAGeCK (FDR < 0.05). For RNA-seq, DESeq2 was used for differential expression (Wald test; FDR < 0.05 and |log₂FC| > 1) and Mfuzz for time-series clustering. Functional enrichment was determined by g:Profiler (g:SCS p < 0.05). All analyses were conducted in the R environment.

3. Results

3.1. Innate Immunity Genes of Camelus Bactrianus

To construct a comprehensive catalog of Bactrian camel innate immunity genes, we integrated multiple data sources. We first compiled a set of 1678 human innate immune genes (HIIGs) from Gene Ontology (GO), InnateDB, and relevant literature [14,15,34]. Using a comparative genomics approach, we then identified camel orthologs by clustering orthologous protein sequences with OrthoVenn3 and the OrthoMCL algorithm, comparing the HIIG sequences (GRCh38) against the Bactrian camel reference sequence (Ca_bactrianus_MBC_1.0). This analysis grouped the 1678 HIIGs into 1366 clusters and the camel proteins into 5798 clusters. We identified 1348 clusters common to both species, yielding a final set of 1535 Bactrian camel innate immune genes (CIIGs) (Figure 1A; Table S1). This refined CIIG set serves as a foundational resource for downstream analyses, including the construction of a targeted CRISPR knockout library and functional studies of the camel innate immune response.

3.2. CRISPR-Cas9 Screening Identified Essential Innate Immunity Genes of Bactrian Camel

Deep sequencing validation of the sgRNA plasmid library prior to screening confirmed excellent quality, with 16,532,184 high-quality reads mapping to 11,742 distinct sgRNAs, achieving 99.99% coverage of the designed library. The even representation of sgRNAs, critical for screening efficacy, was further verified by a homogeneity index of 6.0 (Supplementary Figure S1).

To establish a lethal challenge for the positive selection screen, we first determined that treatment with 50 µg/mL of poly(I:C) or LPS induced near-total cell death in Bactrian camel fibroblasts (Figure 1B). Using this concentration in a genome-wide CRISPR screen (Figure 1C), we identified a limited set of sgRNAs that were significantly enriched in the surviving cell population.

Specifically, under poly(I:C) stimulation, 30 sgRNAs targeting positive selection genes were enriched, while 29 were enriched under LPS stimulation (p < 0.05; Figure 1D; Table S3). Notably, the gene sets conferring survival advantage against viral and bacterial mimics were entirely non-overlapping, indicating distinct genetic requirements for resisting each type of pathogen.

Functional enrichment analysis of these essential genes revealed divergent survival strategies (Figure 1E). We discovered that in the virus mimic knockout library screening group, these genes were primarily enriched in pathways related to positive regulation of response to stimulus, signal transduction, and positive regulation of immune system process. However, in the bacterial mimic knockout library screening group, the genes were predominantly enriched in pathways associated with ubiquitin-protein ligase binding, followed by responses to external stimulus, stress responses, and inflammatory response. These results demonstrate that the survival strategies enabling Bactrian camel fibroblasts to resist viruses or bacteria and remain viable are markedly distinct.

3.3. Poly(I:C) and LPS Stimulation Induces Dynamic Transcriptomic Changes in Fibroblasts

Transcriptomic profiling of camel fibroblasts revealed extensive and time-dependent gene expression reprogramming in response to poly(I:C) and LPS stimulation. Principal component analysis (PCA) demonstrated clear segregation of samples by treatment and time point, confirming the robustness of the transcriptional responses (Figure 2A). The first two principal components collectively explained 81.6% of the total variance (PC1: 63.5%; PC2: 18.1%), with distinct clustering of stimulated samples away from the untreated controls. Differential expression analysis across four time points revealed distinct temporal dynamics for each stimulus (Figure 2B). The antiviral mimic poly(I:C) elicited a rapid response, with the highest number of differentially expressed genes (DEGs) accumulating within the first 6 h (7759 DEGs in total across all comparisons). In contrast, the antibacterial response to LPS was more prolonged, with a significant number of DEGs (8130 in total) still emerging between 12 and 24 h, suggesting a later wave of gene regulation.

Despite these distinct temporal patterns, a substantial core of 5860 DEGs (72.2% of the total) was shared between the antiviral and antibacterial responses, indicating a common foundational immune program (Figure 2C).

To further understand the dynamic changes at the gene expression level during the innate immunity response progression, we classified DEGs into six patterns (Cluster 1, …, Cluster 6) using heatmap and Mfuzz R package for two groups (Figure 3A,B). In the poly(I:C) group, genes in clusters 1, 3, and 6 exhibited up-regulation at 6 h, while genes in clusters 2 and 4 displayed sharp down-regulation. Notably, genes in cluster 5 showed significant upregulation after 12 h, whereas before, these genes had maintained steady expression. The core genes for each cluster are presented to the right of the gene expression cluster plot, alongside the KEGG pathways with significantly enriched genes for each cluster. In the LPS group, genes in clusters 2, 4, and 6 exhibited upregulation at 6 h. Genes in cluster 1 were sharply down-regulated at 6 h, followed by a steady increase. Conversely, genes in cluster 5 showed a sharp down-regulation at 6 h and continued to steadily decrease in expression. The expression pattern of genes in cluster 3 resembled that of cluster 5 in the poly(I:C) group, with expression rising sharply after 12 h. In summary, based on these results, it can be inferred that the progression of the innate immune response is not a gradual and monotonic process but involves nonlinear and drastic transitions at certain time points. This suggests that the gene network connected to key genes is also constantly changing at different time points. We therefore constructed a co-expression network of key genes over time to elucidate the interaction network of innate immune genes.

3.4. Interaction Networks Among Key Genes in the Innate Immunity Process

After analyzing the CRISPR-Cas9 screening of Innate Immunity Genes (IIGs) and gene expression profiles in camel fibroblasts, we identified several pivotal genes involved in innate immunity processes. As depicted in Figure 4A, genes positively selected were expressed in fibroblast cells at four time points. Genes marked with red pentagrams represent those differentially expressed across various temporal groups. Not all positively selected genes were expressed in cellular transcription. For instance, in the antiviral group, genes like CLEC1B and CD300C are crucial for cell survival under lethal concentrations but were not detected at lower concentrations during innate immune responses. In the antibacterial group, unexpressed genes included CXCL10, LOC105065812, and BRS3, illustrating the disparity between genomic mutations and gene expression. However, more importantly, certain genes were found to be pivotal both in genomic knockout experiments and at the transcriptional level, such as HSP90AA1, MMP12, CAV1, BAD, PIAS1, PLA2R1, SOS1, LOC105072346, LOC105071281, NLRP12, and FCGR1A in antiviral group or CSF1, TRIM28, SQSRM1, STX11, TLR1, PRKAR1B, PLEC, NFKBIB, DDX4, DDIT3, and TPP2 in antibacterial group. Especially noteworthy is that HSP90AA1 was significantly enriched among the positively selected genes in the genome knockout study (p < 0.05) and also exhibited notable differential expression in the antiviral transcriptional profile (FDR < 0.05, |log₂FC| > 1)). This gene demonstrated a robust correlation (r > 0.98) at four time points, forming distinct gene networks, as shown in the upper part of Figure 4B. This indicates that the gene networks co-expressed with this gene varied at different immune response time points, implying a dynamic and temporal specificity in gene network interactions. A similar scenario was observed for the CSF1 in the antibacterial group. Indeed, across different transcriptional groups, the overlap of genes with a correlation greater than 0.8 did not exceed 0.67% and 0.43% in antiviral and antibacterial groups, respectively (Figure 4C).

4. Discussion

As a window into the study of the innate immune system, lists of innate immunity genes have been established in multiple species, including humans [17,38], silkworms [19], and oysters [18]. For the first time, we have constructed a collection of innate immunity genes for the Bactrian camel using manual screening and comparative genomics approaches. In cross-species functional gene research, orthologous genes (direct lineage homologs) represent the most rapid avenue for knowledge transfer and learning. This is because orthologous genes, derived from a common ancestral gene and preserved across different species, often retain similar functions, facilitating the extrapolation of genetic and functional insights from one species to another [50]. While this gene collection may not be the most comprehensive, it includes significant gene families such as the Toll-like receptor (TLR) family, NOD-like receptor (NLR) family, heat shock protein (HSP) family, and the tripartite motif (TRIM) family. These families represent major components of the innate immune functional gene research framework, providing a solid foundation for further exploration and understanding of the Bactrian camel’s innate immune system. It should be noted that this study utilized cells from a single female Bactrian camel donor. While this provides internal consistency, future studies should include cells from multiple individuals of both sexes to assess potential individual and sex-specific variations in innate immune responses.

Subsequently, leveraging the constructed innate immunity gene list, we designed a knockout library screening experiment. Fibroblasts are undoubtedly the most convenient and suitable cell model for studying innate immune responses [51,52]. We constructed a knockout cell library of 1535 genes in fibroblasts for the screening of key functional genes in antiviral and antibacterial immune responses. During the positive selection phase of the cell’s antiviral and antibacterial defenses, interference (mutations [42]) in a total of 59 genes enabled the cells to resist stimuli at lethal concentrations. However, the role these genes play at the transcriptional level piqued our curiosity.

To reinforce the robustness and accuracy of our screening results, we undertook a transcriptomic analysis on fibroblasts exposed to stimuli from polyinosinic:polycytidylic acid (poly(I:C)) and lipopolysaccharides (LPS) [53,54,55]. This methodology provided an in-depth view of the fibroblasts’ transcriptional reactions to these specific pathogen-associated molecular patterns (PAMPs), which are recognized mimics of viral and bacterial invasions, respectively. Within the differentially expressed genes, we discerned distinct temporal gene expression clusters, each associated with unique Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways, signifying a multifaceted response to pathogen incursion. Notably, a high level of concordance (exceeding 72%) in differential gene expression was observed across the two immune defense responses, potentially indicating a foundational, non-discriminatory immune response activated by cells in reaction to pathogenic stimuli. Nonetheless, the remaining 28% of gene expressions were specific to the respective stimuli. For instance, gene enrichment in the Ribosome KEGG pathway was a commonality across both stimulus groups. This correlation supports the findings of Christopher Bianco and Ian Mohr [56], who reported that ribosome biogenesis modulates innate immune responses to viral infections and DNA challenges. Our research contributes additional depth to this understanding by demonstrating the pivotal role of the Ribosome pathway in innate immune responses to bacterial stimuli as well, underscoring the complexity and adaptability of cellular immune mechanisms.

Our time-series analysis further delineated the distinct kinetic architectures of the antiviral and antibacterial responses (Figure 3). In fibroblasts challenged with the viral mimic poly(I:C), the transcriptional reprogramming was swift and acute. A substantial wave of gene induction and repression occurred within the first 6 h, encompassing canonical antiviral and stress-response pathways. In stark contrast, the response to bacterial LPS was more protracted and sustained. While an initial transcriptional shift was also observed by 6 h, a significant second wave of differential expression, heavily enriched for genes involved in prolonged inflammatory signaling and tissue remodeling, emerged prominently between 12 and 24 h post-stimulation. This divergence in temporal dynamics—rapid onset versus sustained amplification—underscores that the innate immune system in camel fibroblasts deploys not only stimulus-specific gene sets but also distinct kinetic strategies tailored to the typical persistence patterns of viral and bacterial threats.

By integrating the analyses of the gene knockout library screening and gene transcriptional expression, we discovered that the heat shock protein (HSP) gene family plays a significant role in the immune response to viral infections, particularly the HSP90AA1. Heat shock protein 90 (Hsp90) is a fundamental molecular chaperone that is evolutionarily conserved across various life forms, with its presence confirmed in all kingdoms of life except Archaea [57]. Alterations in Heat shock protein 90 (Hsp90), induced by factors such as mutant forms or heat shock, lead to the release of Heat Shock Factor 1 (HSF1) from its association with Hsp90. This dissociation of HSF1 is a critical event, as it then activates the transcription of antimicrobial peptide genes [58]. Similarly, in studies with wax moths, individuals infected with pathogens and subsequently treated with the Hsp90 inhibitor 17-DMAG exhibited enhanced protection due to an elevated immune response [59]. This implies that inhibition of Hsp90 can fortify the immune system and amplify the defensive response, a finding that aligns with our results. The HSP90AA1 was identified in our knockout library screening and exhibited a rapid increase in expression within 6 h of stimulation by viral mimetics, indicating its involvement in the antiviral immune response. Furthermore, our results also indicate that the expression of this gene at four different time points during viral invasion is associated with a wide array of genes, suggesting that HSP90AA1 may combat viral intrusion by interacting with different gene cohorts at various stages, thereby orchestrating a dynamic and adaptive defense mechanism.

Colony Stimulating Factor-1 (CSF1), also known as macrophage colony-stimulating factor (M-CSF), is a crucial cytokine involved in the regulation of hematopoiesis, particularly in the survival, proliferation, and differentiation of mononuclear phagocyte lineage cells including monocytes, macrophages, and bone marrow progenitor cells [60]. David and colleagues discovered that, using Toll-like receptor 9 (TLR9) as a model for a CSF1-repressed gene, lipopolysaccharide (LPS) induced TLR9 expression in bone marrow macrophages solely in the presence of CSF1. This finding suggests that LPS exposure alleviates the inhibitory effect imposed by CSF1, thereby permitting gene expression [61]. In our results, CSF1 was not only targeted in the knockout library screening stimulated by LPS but also demonstrated a unique transcriptional expression pattern, initially decreasing after 6 h of LPS stimulation and then exhibiting a rapid resurgence at 12 h. This gene was notably detected twice with significant changes in the differential gene expression analysis, corroborating its active engagement in cellular responses. Similar to HSP90AA1, the CSF1 also exhibited a distinct pattern of co-expressed genes at different time points in the transcriptome, with little overlap in the cohorts of genes it interacted with over time. This pattern suggests a dynamic regulatory role for CSF1 in the cellular response to LPS stimulation, adapting its interactions and functions in accordance with the changing cellular environment and response stages.

While our integrated approach provides novel insights into Bactrian camel innate immunity, several methodological limitations should be acknowledged. First, the CRISPR-Cas9 screening in fibroblasts, while experimentally tractable, may not fully recapitulate immune responses in specialized immune cells (e.g., macrophages or dendritic cells). The reliance on a single cell type introduces potential tissue-specific bias, as gene essentiality can vary across cellular contexts. Second, although we employed rigorous sgRNA design criteria (including off-target scoring via CRISPR scan), the pooled screening approach carries inherent limitations: (1) incomplete knockout efficiency, which may obscure weak phenotype-genotype associations; (2) potential confounding by neighboring gene effects in densely packed genomic regions. Third, the orthogonal integration of screening with transcriptomics, while powerful, presents interpretative nuances; for instance, phenotypic hits without expression changes may involve post-transcriptional regulation. Finally, as a foundational study, our work utilized a well-controlled cellular model from a single donor to minimize initial confounding variability. This focused design provides a robust internal reference but underscores the importance of future studies to explore the influence of individual genetic background, sex, and additional cell types on the innate immune landscape we have mapped. These limitations notwithstanding, our findings establish a framework for camel immunogenomics. Future studies could employ single-cell CRISPR screening in primary immune cells, incorporate kinetic transcriptomics at finer resolutions, comparative immunogenomics across uniquely adapted species, and integrate epigenetic profiling to resolve stimulus-specific regulatory logic.

In summary, our study provides a comprehensive analysis of the innate immune response, integrating both CRISPR/Cas9 screening and transcriptomic analysis. The transcriptomic analysis offered insights into the temporal dynamics of immune gene expression and the construction of expression networks for key genes. We specifically highlighted the roles of HSP90AA1 and CSF1 as innate immune genes that play significant roles both genomic and transcriptomic, thereby illustrating the complex interplay of these genes in orchestrating the innate immune response.

5. Conclusions

In conclusion, this study defines the innate immune landscape of Bactrian camel fibroblasts by integrating CRISPR-Cas9 knockout screening and time-resolved transcriptomics. We identified distinct sets of genes essential for survival under lethal viral (poly(I:C)) or bacterial (LPS) mimic challenge. Transcriptomic analysis revealed a shared core response alongside stimulus-specific regulation, and temporal clustering uncovered dynamic gene expression patterns. Through integrated analysis, HSP90AA1 and CSF1 emerged as key drivers, playing critical roles at both functional and transcriptional levels during antiviral and antibacterial responses, respectively. This work provides a foundational resource and candidate genes for understanding immune adaptation in camels.